ELSEVIER Thin Solid Films 259 (1995) 18-24

Structural and optical properties of r.f.-sputtered Si, C, _ x :0 films

E. Ech-chamikh”, E. L. Amezianea, A. Bennounaa, M. Azizan”, T. A. Nguyen Tanb, T. Lopez-Riosb

“Laboratoire de Physique des Solides et des Couches Minces, Facultt! des Sciences Semlalia, Universite Cadi Ayyad, BP SIS, Marrakech, Morocco

bLaboratoire d’Etude des Proprit’tPs Electroniques des Solides, BP 166, 38042 Grenoble Cedex 9, France

Received 26 May 1994; accepted 28 October 1994

Abstract

Oxygenated silicon-carbon alloy (Si, C, _ x :0) films have been prepared by reactive r.f. sputtering. These films were deposited, in a gas mixture of argon and oxygen, from a composite target consisting of a silicon disc on which graphite chips had been placed. The structural and optical properties have been investigated using infrared (IR) spectroscopy, Raman spectroscopy, X-ray photoelectron spectroscopy (XPS) and visible-near-infrared optical absorption. XPS as well as IR measurements show that the films are essentially composed of silicon, carbon and oxygen as expected. The atomic fractions of these elements, deduced from quantitative XPS analysis, are of the order of 50%, 28% and 22% for Si, C and 0 atoms respectively. A study of the chemical shifts of C 1s and Si 2p core level spectra reveals that the films contain not only Si atoms bound to C and 0 but also free silicon and free carbon clusters. The Raman spectroscopy results confirm the presence of free silicon and indicate that the films are amorphous. The optical gap of these films is around 1.3 eV and the refractive index is found to follow the Sellmeier law with a value of about 2.05 at Jti = 2 km.

Keywords: Alloys; Optical properties; Sputtering; X-ray photoelectron spectroscopy

1. Introduction

Amorphous alloys based on silicon and carbon, espe-

cially a-Si, C, _-x and a-Si,C, _-x :H, have been exten- sively studied in the last few years [ 1 - 161. These alloys present some interesting physical and chemical proper-

ties: in particular, they can have a high hardness [ 1, 21 and can be used as window layers in amorphous silicon

solar cells [3, 41. The techniques commonly used to

produce these alloys are: glow discharge [5-91, sputter- ing [ 10-141, evaporation [ 151 and chemical vapour deposition [ 161. For each technique, different starting products can be used. In particular, in the case of the sputtering technique, these alloys can be obtained from a silicon carbide target [lo], from a Si target in a plasma containing a hydrocarbon gas [ 11, 121 or from a composite target [ 13, 141.

In the present work, we have studied the structural and optical properties of Si,C, _,:O films deposited by reactive r.f. sputtering from a composite target.

0040-6090/95/$9.50 0 1995 - Elsevier Science S.A. All rights reserved SSDI 0040-6090( 94)06426- 1

2. Experimental details

Oxygenated silicon -carbon alloy (Si, C, _ x :0) films

have been deposited in an Alcatel SCM 451 r.f. sputter- ing system, which has been described in previous works

[ 17, 1X]. These films were deposited, in a gas mixture of argon and oxygen, from a composite target consisting

of a high-purity silicon target (5N) on which graphite chips had been placed. These chips were uniformly spaced, and covered 50% of the silicon disc surface. The r.f. power and the partial pressures of argon and oxy- gen during the deposition were 250 W, lop2 mbar and lo-’ mbar respectively. Before introducing the sputter- ing gas, the deposition chamber was pumped down to 5 x lop7 mbar using an oil diffusion pump.

The Si, C, _ x :0 films were deposited on the polished face of crystalline silicon substrates for IR, XPS and Raman measurements and on Corning glass (5079) for the visible-near-infrared (V-NIR) optical absorption measurements. A very good estimation of the films’

E. Ech-chamikh et al. I Thin Solid Films 259 (1995) IX-24 19

thickness was obtained by interferential microscopy measurements. All the films used in this study have a thickness of the order of 1.7 urn.

The IR absorption measurements were taken using a double-beam spectrometer (Perkin-Elmer 580B) in the wavelength range between 2.5 and 25 urn. These measure- ments were made with an uncovered substrate (identical to those on which the films were deposited) in the reference compartment.

The Raman spectroscopy measurements were carried out using a Jobin Yvon UlOOO spectrophotometer. The line 530.9 nm of a krypton laser, with a power of 80 mW, was used as the excitation source.

The XPS analyses were preformed on a VSW (Vacuum Science Workshop) surface analysis apparatus fitted out with a hemispherical analyser. All the XPS spectra were taken in the FAT (fixed analysis energy transmission) mode and Mg Kx radiation was used as the excitation source.

The optical transmission in the V-NIR wavelength range (between 0.4 and 2.5 urn) was measured using a double-beam spectrophotometer (Shimadzu UV-3101 PC).

3. Results and discussion

3.1. Structural and chemical analysis

3. I. I. IR spectroscopy Fig. 1 shows the IR spectrum (curve a), obtained on

a Si,C, _ .:O film, represented as the inverse of the transmittance, l/T, versus the wavenumber, l/L This spectrum shows the presence of three main absorption bands, which are located at 1060, 790 and 460cm-‘. These bands are attributed to the stretching vibration modes of Si-0 [ 19, 201 and Si-C [ 14, 211 bonds and the bending mode of Sii0 bonds [ 19, 201 respectively. The band located at 1060 cm-’ shows a shoulder around 1200 cm-‘, which can be attributed to the stretching mode of C-O bonds [22]. The spectrum also shows interference fringes due to multiple reflections in the film. These fringes present some irregularities, which are due to the existence of other weak absorption bands. In order to extract these bands we have eliminated the interference fringes by dividing 1 /T spectrum by a sinusoidal function as A + B cos(C/;I). In this procedure, the constants A, B and Cwere adjusted so as to fit perfectly the regions where there are no absorption bands (Fig. 1, curve b). In fact, the inverse of the transmittance for a weakly absorbing film can be written as follows [23]:

-f = exp(ud)[A + B cos(4nnd/2)] (1)

where a, d and n are the absorption coefficient, the thickness and the refractive index of the film respec-

6 . . . . . . . . . . . . . . . . . . .

,p! _.__...... .&_A.+ ot.‘..‘..‘.‘..‘.‘....i

0 1000 2000 3000 4000

WAVENUMBER (cm-‘)

Fig. 1. Infrared absorption spectrum (-) and the sinusoidal fit of

the interference fringes as explained in the text (- - -).

m; 12, . . . , , . . . , . . . ,

‘! c

Si-C (s)

Arom. C-C (5); O-H (b)

0 1000 2000 3000 4000

WAVENUMBER (cm-’ )

Fig. 2. Absorption coefficient versus the wavenumber as obtained

after the elimination of the interference fringes.

tively, and A and B are constants depending on the refractive indices of film and substrate. The term [A + B cos(4nnd/l)] represents the inverse transmit- tance of a hypothetical transparent film having an optical thickness nd. It can be considered as the baseline to get CI. This procedure has been already applied successfully to other materials [24, 251. The best fit shown in Fig. 1 has been obtained with an optical thickness nd equal to 3.745 urn. If we take the value of the refractive index deduced from the NIR region, which is about 2.05 (see Section 3.2), the corresponding thickness is about 1.8 urn. This value is close to that obtained from the V-NIR absorption measurements (Section 3.2). The difference between these two thick- ness values can be due to both the uncertainty in the n value and the thickness inhomogeneity effect [26].

Fig. 2 shows the plot of c1 versus the wavenumber l//1 as obtained from the method described above. In this spectrum we can see not only the principal bands identified above but also some other weak absorption bands located around 3400, 2900 and 2100 cm-‘, which

E. Ech-chamikh et al. I Thin Solid Films 259 (1995) 18-24 20

? 9

: M

1

200 600 1000 1400 1800 2200

RAMAN SHxFr b“)

Fig. 3. Raman spectra obtained on (a) a Si,C, _,:O film, (b) crys-

talline silicon substrate and (c) amorphous carbon film.

are due to the O-H [22], C-H [27, 281 and Si-H [29, 301 stretching modes respectively. Moreover, there is a wide structure between 1250 and 1850 cm-‘, which may be associated with the superposition of several bands. These bands can be due to the stretch- ing mode of C=O bonds centred at 1700 cm-’ [27], the bending mode of O-H bonds around 1600 [22] and the stretching modes of C-C and C=C bonds in different configurations between 1250 and 1650 cm- ’ [31]. The presence of O-H, C-H and Si-H bonds shows that the films are slightly polluted by hydrogen, which probably originates from water vapour ab- sorbed during the air exposure (a few days) before their analysis.

3.1.2. Raman spectroscopy Fig. 3 shows the Raman spectrum obtained on a

Si, C, _-x :0 film deposited on a crystalline silicon sub- strate (curve a). For comparison, the Raman spectra of the silicon substrate and a 0.3 urn thick amorphous carbon (a-C) film are also shown in the same figure (curves b and c respectively). The S&C, _-x :0 spectrum shows a wide peak centred around 480 cm-‘, which is shifted by 40 cm-’ towards low frequencies from the crystalline silicon one. This wide peak is due to Si atoms bound to other ones in an amorphous state [ 13, 321. This result suggests that the Si,C, _,:O films contain free silicon clusters, as confirmed also by XPS results. Moreover, we note also the presence of a weak and wide structure between 600 and 1000 cm-‘, which can be attributed to Si-C bonds in amorphous silicon carbide [ 131. Finally, the Si, C, _-x :0 spectrum does not show the wide band characteristic of a-C, which appears usually between 1200 and 1700 cm-’ (curve c), attributed to bonds between C atoms in the sp2 configuration [32]. This result suggests that the free carbon clusters observed by XPS (Section 3.1.3) are essentially in the sp3 configuration.

I”“““““““““““““‘I 01s

I....,....‘..,....~.

0 200 400 600

BINDING ENERGY (eV)

Fig. 4. Widescan XPS spectra obtained (a) before and after two

successive ionic argon bombardments of (b) 60 min and (c) 45 min.

3.1.3. XPS spectroscopy Fig. 4 shows the XPS spectra obtained on a

Si,C, _-x :0 film before (spectrum a) and after two successive ionic argon (A?) sputter etchings during 60 and 45 mn (spectra b and c respectively). The spec- trum obtained before etching shows four peaks located at 100, 155, 285 and 532 eV in binding energies. These peaks are due to the Si 2p, Si 2s C 1s and 0 1s core levels respectively. After the first etching (spectrum b), the intensities of the Si peaks increase while the 0 1s one decreases considerably. This result shows that the film surface contains an oxygen excess, which originates from the absorption of atmospheric oxygen and/or water vapour during the air exposure (a few days) before the introduction of the film into the analysis chamber. Moreover, after this first etching, an addi- tional small peak appears at 242 eV. The latter, which is attributed to the Ar 2p core level, is due to argon atoms incorporated in the film during the Ar+ sputter etching. The spectrum obtained after the second etching (spec- trum c) is similar to that obtained after the first one with a relatively small decrease of the 0 1s peak intensity and a slight increase of the C Is, Si 2p and Si 2s peak intensities. So, spectrum c can be considered as representative of the bulk composition of the Si, C, _ x :0 films.

In XPS, the integrated intensity, IA, of the signal originating from a given core level of an element A is proportional to the atomic concentration, NA, of this element [ 331:

IA = a,&, TN, (2)

where T is the analyser transmission, gA is the ioniza- tion cross-section of the considered level and 2, is the inelastic mean free path (IMFP) of the photoelectrons originating from this level. For photoelectrons having a kinetic energy EA exceeding 300 eV (which is in partic- ular verified for the 0 Is, C 1s and Si 2p photoelectrons excited by the Mg Kcc line) the IMFP is proportional to

E. Ech-chamikh et al. I Thin Solid Films 259 (1995) 18-24 21

96 98 100 102 104 106 200 282 284 286 208 290

BINDING BNBRGY (eV) BINDING ENERGY (eV)

Fig. 5. Si 2p core level spectra obtained before argon sputter etching (curve b) and after the second etching (curve c): ,A), experimental spectra;

- - -, deconvoluted gaussian components; -, envelopes (sum of the components). Spectrum of pure Si (curve a) is also given for comparison.

The energy positions of Si 2p in pure Si, Sic, SiO,,C,, and SiO, are indicated by the arrows.

Fig. 6. C 1s core level spectra obtained before argon sputter etching (curve b) and after the second etching (curve c): iI, experimental spectra;

- - -1 the deconvoluted gaussian components; -, the envelopes (sum of the components). Spectrum of pure a-C sample is also given for

comparison. The arrows indicate the C Is energy positions in different atomic configurations.

the square root of EA [34]. On the other hand, T varies as l/EAIIz because the XPS spectra were taken with the FAT mode [35]. So, the atomic concentrations of C and 0 atoms relative to the Si one are given by

and

(3)

(4)

Taking into account the ionization cross-sections (rela- tive to the C 1s core level) of both 0 Is, C 1s and Si 2p levels, which are 2.85, 1 and 0.9 respectively [36], we have calculated the No/N,, and Nc/Nsi ratios using Eqs. (3) and (4). The values obtained are 0.43 and 0.56 respectively. The corresponding atomic fractions of Si, C and 0 atoms are 50%, 28% and 22% respectively.

In order to study the chemical environments of Si and C atoms in the films, we have recorded the XPS spectra in the Si 2p and C 1s peak regions with a relatively slow scan rate. The spectra obtained after integral background subtraction are shown in Figs. 5 and 6. In these figures, we report also the Si 2p and C 1s spectra obtained on amorphous silicon and carbon films (used as references) respectively.

The Si 2p core level spectrum obtained before ion etching (Fig. 5, spectrum b) shows two overlapping peaks located around 100.4 and 103.3 eV. These peaks

are shifted by 0.9 and 3.8 eV towards higher binding energies in comparison with the Si 2p peak obtained on the pure silicon film (Fig. 5, spectrum a). These chemi- cal shifts can be attributed to Si atoms in silicon carbide [37] and silicon dioxide [24] respectively. After the two successive etchings, the intensity of the peak located around 100.4eV increases while the second peak (lo- cated at 103.3 eV) reduces to a shoulder (Fig. 5, spec- trum c). This result is in agreement with the surface oxygen excess mentioned above and shows that the Si atoms are essentially bound to C atoms as SIC. To separate the contributions of the different Si atoms’ chemical environments, we have deconvoluted the experimental Si 2p spectra using four gaussian components. In fact, in addition to the SIC and SiO, contributors mentioned above, two other components had to be taken into consideration: the first, which is expected to be due to Si atoms in free silicon clusters detected by Raman spectroscopy (see Section 3.1.2) must be located at 99.5 eV, and the second is expected to be due to Si atoms in suboxide (SiO, < 2) and SiO, C, configurations. In this deconvolution procedure, the full width at half maximum (FWHM) and the position of the pure Si component were fixed equal to 1.6 eV and 99.5 eV respectively according to the spectrum obtained on pure Si film (Fig. 5, spectrum a). The component parameters that allow the best fit of the experimental spectra (as shown in Fig. 5) are given in Table 1 together with the integrated intensities at the surface

22 E. Ech-chamikh et al. / Thin Solid Films 259 (1995) 18-24

Table I Positions and FWHMs of the gaussian components that allow the best

fit of the Si 2p core level spectra as shown in Fig. 5. The corresponding

integrated intensities are also reported

Components Position (eV) FWHM (eV) Integrated intensities

in surface in bulk

Pure Si

Sic

SiO., C

SiO, ’

99.5 1.6 0.03 0.70

100.4 I.8 2.72 5.88

101.5 2 1.97 2.84

103.2 2 3.15 1.53

(before ion etching) and in the bulk (after ion etching). Taking into account the fact that these intensities are proportional to the concentration of Si atoms in the corresponding configurations, we can deduce that 54% of Si atoms are bound to C ones as Sic, 14% are bound to oxygen as SiOz, 6% are bound exclusively to Si atoms (silicon clusters) and 26% are in the suboxide and SiO, C, configurations.

The C 1s spectrum obtained before ion etching (Fig. 6, spectrum b) shows also two overlapping peaks lo- cated around 283.2 and 284.9 eV. After surface etching, the intensity of the first peak (283.2 eV) increases while the second one reduces to a weak shoulder (Fig. 6, spectrum c). The main peak (283.2 eV) is shifted by 1.3 eV towards lower binding energies in comparison with the C 1s peak obtained on the a-C film (Fig. 6, spectrum a). Such a chemical shift can be attributed to C atoms in SIC [37]. In order to separate again the contributions of the different atomic environments, we have deconvoluted the C 1s experimental spectra using gaussian components. For the spectrum representing the bulk (spectrum c), the best fit is obtained using three components, which are expected to be due to C atoms in SIC configuration (mentioned above), to car- bon clusters and to C atoms bound to oxygen. In this deconvolution procedure, the FWHM and the position of the free carbon component were fixed equal to 1.75 eV and 284.5 eV respectively as deduced from the C 1s peak obtained on the pure a-C film (Fig. 6, spectrum a). The FWHMs and the positions of the two other components are found to be around 1.55 eV and 283.2 eV for the first and 1.8 eV and 285.8 eV for the second respectively. The first component (283.2 eV) has already been identified above as originating from C atoms in Sic. The second one (285.8 eV), shifted by 1.3 eV towards higher binding energies in comparison with the C 1s peak obtained on the pure a-C film, can be attributed to C atoms bound to oxygen as C-O [24, 381. Taking into account the integrated intensities of these components, we have estimated the fractions of C atoms in the three different configurations cited above as we have done for Si atoms. These fractions are

’ B ” ” ” ” ” ” ”

0.4 0.8 1.2 1.6 2 2.4

WAVELENGTH ( pm)

Fig. 7. V-NIR transmission spectrum obtained on a Si, C, _ x :0 film.

71% for Sic, 25% for free carbon clusters and only 4% for C-O. In comparison with the fraction of Si atoms bound to oxygen (mentioned above), this result indi- cates that the oxygen atoms are preferentially bound to silicon. For the C 1s spectrum obtained before etching, a good fit requires not only the three components used above but also a fourth additional component located around 287.2 eV (Fig. 6, spectrum b). The latter com- ponent having a FWHM of 2 eV is due to C=O bonds [24, 381. This result is in agreement with the oxygen excess at the surface of the films mentioned above.

3.2. Optical properties

Fig. 7 shows the V-NIR trasmission spectrum of a Si,C, _-x :0 film. From this spectrum, which presents interference fringes due to multiple reflections in the film, we deduced a thickness value of 1.72 pm and the variations of both the refractive index n and the absorp- tion coefficient a with wavelength (Figs. 8 and 9 respec- tively). Details concerning the determination of these parameters from the transmission spectrum, using the interference fringes, have been reported elsewhere [ 391.

Between 0.8 and 2.4 pm, the refractive index varies slightly (by about 6%) and follows the Sellmeier law [40] :

n ‘(4 = %I* + A2

[ 1 (n 2 _ &*)

The values of the constants n,, 1, and &, that give the best fit of the experimental n(L), as shown in Fig. 8, are 2.04, 0.59 pm and 0.0012 pm respectively.

The optical gap value &, (defined as the energy at which CI is equal to lo4 cm-’ [41, 421) can be obtained from the semi-logarithmic plot of o! versus photon energy E (Fig. 9). This value is 1.84 eV.

In many amorphous semiconductors, the absorption edge can be divided into two regions: above and below a z lo4 cm-‘.

E. Ech-chamikh et al. I Thin Solid Films 259 1199% IS-24 23

2.25

--=---A - EJ

0.5 1 1.5 2 2.5 z,E,""",",3 1 1.5 3

WAVELENGTH ( Pm) PHOTON ENERGY (eV)

Fig, 8. Plot of the refractive index versus wavelength as deduced from the VNIR transmission spectrum: 0, experimental values; ~~ , Sellmeier

law tit.

Fig. 9. Semi-logarithmic plot of the absorption coefficient G( versus the photon energy E.

e CI

G d

Y

300LI I I , , , , , , , , , , , , , , , ‘3

-0.5 1 1.5 2 2.5

E (eV)

Fig. IO. Results of Fig. 9 replotted as (nE)“’ versus E.

For values lower than lo4 cm-‘, CI follows an expo- nential law [43, 441:

c( = cio exp (6)

where LYE and E, are constants and E, is the Urbach energy characterizing the disorder. This behaviour is well verified in the case of our films (as can be seen in Fig. 9) with an E, value of about 350 meV. This value, which is greater than those obtained in the case of a-Si:H films with a small content of defects (50- 100 meV) [45-471, indicates that our films present a relatively high defect density.

For values greater than lo4 cm-‘, SI follows the Taut’s law [45]:

ctE = A(E - ET)2 (7)

where A is a constant and ET is the Taut’s optical gap. Fig. 10 shows the (aE) ‘I* versus E plot. For a values greater than lo4 cm-’ (i.e. E greater than EM), the obtained curve is a straight line according to Eq. (7).

The extrapolation of this linear portion (at c( = 0) gives an ET value of 1.31 eV. This value is closer to Eo3 (the energy at which CI is equal to lo3 cm-‘) than to Eo4.

This result is in agreement with Mott’s proposition [46]; he pointed out that, in some amorphous materials, the variation of the optical gap with temperature may lead to its position near E,,,.

4. Conclusion

We have studied both the structural and optical properties of r.f.-sputtered Si, Cl _-\ :0 films using IR, XPS, Raman and V-NIR optical absorption tech- niques.

These films are found to be amorphous, and contain not only silicon atoms bound to carbon and oxygen (in Sic, SiO, and SiO,C configurations) but also a rela- tively small content of free silicon and free carbon clusters. The atomic fractions of Si, C and 0 atoms are of the order of 50%, 28%’ and 22% respectively.

The Taut’s optical gap of these films is of the order of 1.3 eV and the refractive index follows the Sellmeier law with a value of about 2.05 at A = 2 pm.

References

[I] M. A. Bayne, Z. Kurokawa, N. U. Okorie, B. D. Rot,

Johnson and R. W. Hess, Thin Solid Films, 107 (1983) 201.

[2] H. Yoshihara, H. Mori and M. Kiuchi, Thin Solid Films.

(1981) I.

L.

76

[3] Y. Tawada, H. Okamoto and Y. Hamakawa, Appl. Phys. Left.,

32 (1981) 237.

[4] Y. Tawada, M. Kondo, H. Okamoto and Y. Hamakawa, Sol.

Energy Mater., 6 ( 1982) 299.

[5] D. R. McKenzie, N. Savvides, D. R. Mills, R. C. McPhedran

and L. C. Botten, Sol. Energy Mater., Y ( 1983) 113.

24 E. Ech-chamikh et al. 1 Thin Solid Films 259 (1995) 18-24

[6] A. H. Mahan, B. V. Roedern, D. L. Williamson and A. Madan,

J. Appl. Phys., 57 (1985) 2717. [7] S. Ray, D. Das and A. K. Barua, Sol. Energy Mater., 15 (1987)

45. [8] D. K. Basa and F. W. Smith, Thin Solid Films, 192 (1990) 121. [9] T. Takeshita, Y. Kurata and S. Hasegawa, J. Appl. Phys., 71

(1992) 5395. [IO] K. Wasa, T. Nagai and S. Hayakawa, Thin Solid Films, 31

(1976) 235. [ 111 A. Guivarc’h, J. Richard, M. Le Contellec, E. Ligeon and J.

Fontenille, J. Appl. Phys., 51 (1980) 2167. [ 121 A. J. Learn and K. E. Haq, Appl. Phys. Lett., 17 (1970) 26. [13] A. Morimoto, T. Kataoka, M. Kumeda and T. Shimisu, Phil.

Mag. B, 50 (1984) 517.

[14] K. Yoshii, Y. Suzaki, A. Takeuchi, K. Yasutake and H.

Kawabe, Thin Solid Films, 199 (1991) 85. [15] A. Boyer and D. Deschacht, Le Vide les Couches Minces, 221

(1984) 135.

[16] R. M. Young and W. D. Partlow, Thin Solid Films, 213 (1992) 170.

[17] M. Khaidar, A. Essafti, A. Bennouna and E. L. Ameziane, J. Appl. Phys., 65 (1989) 3238.

[18] A. Bennouna, E. L. Ameziane, A. Haouni, N. Ghermani, M.

Azizan and M. Brunel, Sol. Energy Mater., 20 (1990) 405. [ 191 N. Kniffper, B. Scroder and J. Geiger, J. Non Cryst. Sol., 58

(1981) 153.

[20] A. Cachard, J. A. Rocher, J. Pivot and G. H. S. Pupuy, Phys. Stat. Sol. (a), 5 (1971) 637.

[21] M. Nishikuni, K. Ninomiya, S. Okamoto, T. Takahama, S.

Tsuda, M. Ohnishi, S. Nakano and Y. Kuwano, Jpn. J. Appl. Phys., 30 (1991) 7.

[22] R. A. Nyquist, in J. W. Robinson (ed.), Handbook of Spec- troscopy, Vol. II, CRC Press, Boca Raton, FL, 2nd ed., 1979, p.

68.

[23] 0. S. Heavens, Optical Properties of Thin Solid Films, Dover,

New York, 1965.

[24] E. Ech-chamikh, M. Azizan, E. L. Ameziane, A. Bennouna, M.

Brunel and T. T. A. Nguyen, Sol. Energy Mater. Solar Cells, 31 ( 1993) 187.

[25] E. Ech-chamikh, E. L. Ameziane, A. Bennouna, M. Azizan, Y.

Bounouh, M. L. Theye, J. Cernogora and J. L. Fave, Sol. Energy Mater. Sol. Cells, 33 (1994) 361.

[26] M. Benmoussa, A. Bennouna, E. Ibnouelghazi and E. L. Amezi-

ane, Sci. Tech. Appl. du Vide, 275 (1995) in press.

[27] D. R. McKenzie, R. C. McFedran, N. Savvides and L. L.

Batten, Phil. Mag. B, 48 (1983) 341. (281 P. Gouderc and Y. Catherine, Thin Solid Films, 146 (1987)

93. [29] E. C. Freeman and W. Paul, Phys. Rev. B, 18 (1978) 4288. [30] D. R. McKenzie, J. Phys. II, 18 (1985) 1935.

[31] J. Robertson, Progr. Solid State Chem., 21 (1991) 199.

[ 321 M. Yoshikawa, Mater. Sci. Forum, 52/53 (1989) 365.

[33] L. C. Feldman and J. W. Mayer, Fundamentals of Surface and Interface Analysis, North-Holland, Amsterdam, 1986, p. 213.

[34] C. R. Brundle, Sur. Sci., 48 ( 1975) 99.

[35] D. B. B. Lollman, Thesis, Universite Grenoble I, France, 1992.

[36] J. H. Scofield, J. Electron Spectrosc Relat. Phenom., 8 (1976) 129.

[37] E. Ech-chamikh, M. Azizan, E. L. Ameziane, J. Y. Veuillen and

M. Brunel, Sol. Energy Mater. Sol. Cells, 29 ( 1993) 131.

[38] J. F. Moulder, J. S. Hammond and K. L. Smith, Appl. Surf Sci., 2.5 (1986) 446.

[39] A. Bennouna, Y. Laaziz and M. A. Idrissi, Thin Solid Films, 213 (1992) 55.

[40] G. Bruhat, in A. Kastler (ed.), Cours de Physique GPnPrale -Optique, Masson & Cie, Paris, 5th ed., 1959, p. 372.

[41] J. Stuke, J. Non Cryst. Solids, 4 (1970) 1.

[42] E. C. Freeman and W. Paul, Phys. Rev. B, 20 (1979) 716. [43] F. Urbach, Phys. Rev., 92 (1953) 1324.

[44] G. D. Cody, T. Tiedja, B. Abeles, B. Brooks and Y. Goldstein,

Phys. Rev. Lett., 47(1981) 1480. [45] J. Taut and A. Menth, J. Non Cryst. Solids, 8-10 (1972)

569. [46] N. F. Mott and E. A. Davis, Electronic Processes in Non

Crystalline Materials, Clarendon Press, Oxford, 2nd edn., 1979,

p. 283.

[47] Y. Laaziz and A. Bennouna, Sol. Energy Mater. Sol. Cells, 31 (1993) 23.